Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I trained an AI model on 120M+ songs from iTunes

maroofy.com

Post reply on HN
Print view View original

371–380 of 444 posts

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#371

Post by raffraffraff »

macrolime wrote: post #352
raffraffraff wrote: post #204

I'm wondering if anyone has done something similar, but instead of trying to find similarities in the raw audio, they use tags available from sources like Last.FM, Musicbrainz, Discogs etc? And the ultimate answer to that is probably "those sources kinda suck". Discogs is like a trainspotter on the spectrum, fascinated by release IDs. Musicbrainz is kinda similar (each song will have a dozen matches of wildly differe…

I've tried with Discogs and found it to work pretty well. Kinda similar to what OP did just the "embedding" vectors was created by the Genre/Styles on Discogs. I didn't have a Vector database though, so it was kinda very slow. On Discogs those tags are per album and not per track. To create a playlist of say 10 songs similar to a song, I'd find the ten closest albums, then search for them on last.fm and pick the most…

Per-album metadata is useless for a lot of stuff that I like. It's even useless for a lot of The Beatles stuff because they tend to have a range of styles on an album and tended to bring in weird instruments on individual tracks.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#372

Post by sitkack »

no_butterscotch wrote: post #163
sitkack wrote: post #143

Earlier quoted context omitted.

It chokes if you have any adblocker or anti-tracking.

It chokes on everything: Brave or Chrome, incognito with Brave Shield disabled or uBlock disabled also don't work, same with Safari, ... I'm glad it includes the examples at the bottom so I can see what it does though.

I was only able to get search to work on Chrome in incognito on a mac.

It basically overfits on the similarity, which is kinda neat in a way. It could be extremely useful in scoring video, you want something "like" this song, it gives you lots of things to choose from. Or if I wanted to make a set list on a theme.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#373

Post by emehrkay »

subtech wrote: post #251
bombela wrote: post #190

Everything loads instantly. Plays almost instantly. And it really seems to find very similar style/beat/music. Interface is clean. I am not sure if intentional, but the loading animation on the play buttons feels like it is in sync with when the music starts playing. Makes for a responsive feedback.

Thanks! :) Let's just say that setting up the backend systems involved a lot of tears & frustration lmfao

What tech did you use to pull down all of this data and comb through it?

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#375

Post by Toutouxc »

huhtenberg wrote: post #306

Search doesn't work in a recent Firefox on Windows - typing anything in the search box shows the "Loading..." dropout below it and then nothing happens. This generates no network activity and the console shows: Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at https://cdn.segment.com/v1/projects/F4GFNelOpRsgUJc6iwTuiXr2t6AH5LCY/settings. (Reason: CORS request did not succee…

Getting the same error on macOS, both Safari and Firefox.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#377

Post by pandemicsoul »

dangwhy wrote: post #376

I tried some songs but recommendations were tracks that sampled that song and songs with same name. I am not sure if thats quite what i am looking for.

Yes, came here to say something similar. I searched for a song that was playing on at the moment, and what I got back were six other versions of the same song, at the top of the list, from the same artist from different albums.

Covers of this song would be great! But probably worthwhile to exclude anything with the same name by the same artist.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#378

Post by microtonal »

anagram666 wrote: post #148
knaik94 wrote: post #24

This is very interesting, but unfortunately I haven't had the greatest luck in finding new songs I would enjoy listening to. It absolutely finds similar sounding tracks, but it doesn't distinguish which part of the song made it enjoyable. There's no tempo consistency or genre consistency or even main instrument/vocal timbre consistency between recommendations. I think locking one or more of those dimensions would all…

If I am not mistaken, it this is only trained on the preview and not on the entire song. If you listen to a music with a real intro, it gives strange results. For example: "Goodbye Blue Sky - Pink Floyd" ( https://maroofy.com/songs/1065976153 )

Same for "Station by Station - David Bowie" -- lot's of tracks with ambient noise.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#379

Post by chrisgd »

pvg wrote: post #218
chrisgd wrote: post #165

I find it disheartening that the comments are loaded with people saying how bad it is. The interface is awesome, the ability to have search results pop up and offer suggestions is impressive. I thought the results for a Lana Del Ray song and a song by Carlos Santana were both interesting and suggested new things. We should do a better job of supporting creators here.

a better job of supporting creators here You can do that by commenting on the things you liked about the project! But without the meta because that just begets more meta and ends up detracting from the thing being showhn.

I did the latter and think the former is necessary. It is like when Dropbox was on here and everyone shit on it. You can offer criticism without being a jerk.

Re: Show HN: I trained an AI model on 120M+ songs from iTunes

#380

Post by from »

yreg wrote: post #262
charcircuit wrote: post #52

Earlier quoted context omitted.

The preview audio is free. It may just be 120,000,000 previews.

Even so, how do you get the 120M previews?

Apple has a public API with some rate limiting that returns a link to the preview audio file, see:

https://itunes.apple.com/us/lookup?id=1023678453 (https://www.chrisjmendez.com/2017/06/19/working-with-itunes-...)

So probably all that is required is a couple threads downloading and a proxy service with a large pool of IP addresses randomly rotating on every request. Maybe OP also found an undocumented API endpoint somewhere that was not rate limited.

Post reply on HN